In general, the assumption is that two point sources of light can be resolved (separately imaged) if the centre of the
Airy disk, generated by one of the sources, at most overlaps with the first order reflection in the diffraction pattern of the second Airy disk; a condition known as the Rayleigh Criterion.
The resolution is obviously an important parameter for the selection of a CCD camera. The digital images should show all the fine details revealed by the microscope magnification (unless the experimenter chooses to forsake resolution to some extent by binning for the benefit of other parameters). Digital images consist of pixels which represent the data of the miniature photodiodes on the CCD chip. Each photodiode integrates the intensity of a tiny area of the microscope image. It is easy to understand that the image resolution depends on the total number of pixels it is composed of and correspondingly on the number of photodiodes on a CCD chip of a certain size or, in other words, inversely on the dimensions of the CCD pixels.
To give an example, the Sony ICX285 chip that is utilised in some of the latest generation interline-transfer CCD cameras consists of 1392x1040 pixels of 6.45 x 6.45 microns size. The dimensions of digital image pixels are given by the CCD pixel size divided by the microscope magnification. Image pixels obtained with a 60x objective would thus represent a sample area of about 108 x 108 nm.
The diffraction limited resolution of an epi-fluorescence microscope is given by {0.61 times wavelength divided by objective NA}. So, at 500nm the resolution limit is 218 nm with a 1.4 NA objective. (In "real life" the resolution is usually worse.) With the pixels being half this size in our example above, the camera resolution is just barely enough to match the (theoretical) microscope resolution. According to a rule of thumb (actually derived from the Nyquist criterion), the pixel size should be more than two times or rather three times smaller than the finest spacing that shall be detected.
It should be noted in this discussion, however, that even if small CCD pixels improve the resolution, they also reduce the dynamic range of the device because the full-well capacity depends also on the size of the photodiodes. Furthermore, the average signal per pixel is directly dependent on the size (area) of the pixels.
